Biomedical application of optical fibre sensors

Perhaps the simplest way to achieve direct interaction of light with the medium surrounding the optical fibre is through the removal of the cladding to allow access to the EW, as illustrated in figure 2.

Zoom In Zoom Out Reset image size

A change of the optical properties of the surrounding medium, for example, RI or spectral absorption, will lead to modulation of the light propagating in the optical fibre. This fibre modification facilitates EW spectroscopy [13], which is a highly sensitive and powerful technique that is used to measure quantitatively and qualitatively the chemistry of the environment surrounding the optical fibre. The intensity of the EW decays exponentially with the distance from the interface between the fibre core and the surrounding environment. The penetration depth (d_p) of the EW is described by:

$$\begin{eqnarray}&&{d}_{p}=\displaystyle \frac{\lambda }{2\pi {({n}_{{\rm{e}}{\rm{f}}{\rm{f}}}^{2}-{{n}_{s}}^{2})}^{\mathrm{1/2}}},\end{eqnarray} \tag{ 1 }$$

where λ is the wavelength of light in free space, n_s is the RI of the surrounding environment and n_eff is the effective RI of the mode guided by the optical fibre [10].

The absorption spectrum of the medium surrounding the fibre influences the attenuation of the EW, and thus of the mode, according to the Lambert–Beer law:

$$\begin{eqnarray}&&\displaystyle \frac{I}{{I}_{0}}=c* \alpha * L,\end{eqnarray} \tag{ 2 }$$

where α is the molar extinction coefficient, c is the concentration of the absorbing substance, L is the path-length that light travels through the sample and I_o and I are the light intensities at the input and output of the optical fibre, i.e. before and after interaction with the absorbing substance, respectively.

In its simplest form, coupling the output from a broadband optical source into the proximal end of the fibre and observing the transmitted light by coupling the output from the distal end to a spectrometer will allow the spectroscopy of the surrounding medium to be measured. An example of the use of EW spectroscopy, undertaken using a hard-clad multimode silica optical fibre with a section of the plastic cladding removed, to measure the absorption spectrum of a porphyrin dye compound is shown in figure 3 [13].

Zoom In Zoom Out Reset image size

A variation of this type of OFS involves the creation of a 'U' shaped bend in a section of multimode optical fibre [14]. Bending the fibre causes the EW to extend further into the surrounding medium, increasing the interaction between the EW and the measurand [14]. There is also an implementation advantage as the source and detector are positioned on the same side of the medium under investigation. It is also worth mentioning that the low attenuation of chalcogenide glasses in the infrared (IR) region (1–10 μm), where specific absorption features of organic molecules are located, makes the use of these fibres attractive for EW spectroscopy [15].

Access to the EW can be also gained via tapering of a relatively short section (of length ranging from sub-millimetre to tens of millimetres) of the optical fibre. A tapered optical fibre consists of a region of fibre with reduced and uniform diameter (the waist) that is bounded by conical sections where the diameter of the fibre changes to merge the tapered section with the unperturbed surrounding fibre, as illustrated in figure 4(a). Tapered fibres are fabricated by heating and stretching the optical fibre with the heat source being typically a flame or the output from a CO₂ laser. The optical properties of the tapered fibre waveguide are influenced by the profile of the conical tapering sections, by the diameter of the taper waist and by the RI of the surrounding medium. The proportion of the power in the EW, and thus the interaction with the surrounding medium, increases with decreasing diameter of the taper waist, and with decreasing difference between the RI of the fibre and surrounding medium [16, 17]. Tapered optical fibres offer a number of attractive features for sensor development, including large evanescent fields, flexibility and compactness. In the case of a tapered optical fibre that is coated with a functional material, the properties of the waveguide are influenced by the optical thickness (product of the RI and geometrical thickness) of the coating.

Zoom In Zoom Out Reset image size

Originally, tapered optical fibres were employed for the development of directional couplers, where two or more tapers are fused together, to provide efficient light coupling between fibres [18]. They have also found applications in sensor development [19], light amplifiers [20] and near and far field microscopy [21].

Optical fibre tapers are classified as adiabatic and non-adiabatic. In adiabatic configurations, the angle of the taper transition region is small (10⁻⁴ to 10⁻³ rad [22]) and the cylindrical symmetry of the optical fibre is retained (with taper ratio of taper diameter to initial diameter of the core (a/a₀) between 0.2 and 1 [22]), resulting in most of the optical power remaining in the fundamental mode [16]. The uniform transmission spectrum of adiabatic tapers facilitates their use for EW spectroscopy. In non-adiabatic tapers, the geometry of the taper is such that the fundamental mode is coupled into higher order modes. For non-adiabatic tapers of diameter less than 10 μm, the linearly polarized, LP₀₁, mode of the single mode fibre generally couples to the hybrid, HE₁₁ and HE₁₂, modes of the tapered waist [23]. Interference effects in recombination of the excited modes at the 2nd tapering section introduce channelled spectrum features into the transmission spectrum, the wavelengths of which are sensitive to differential changes in the effective index of the modes, which arises from the different penetration depths of their evanescent fields. This has been exploited to allow measurements of the RI of the surrounding environment or in the RI of chemically sensitive material deposited onto the surface of the tapered section of fibre.

Tapered optical fibres have been exploited as sensor platforms for a variety of applications [24–26]. There are two approaches that are used to exploit the interaction between the EW and the surrounding medium. The first involves the measurement of the attenuation of light propagating through the tapered region, which is influenced by both the spectroscopy and RI of the surrounding medium [15]. The second concerns the influence of the surrounding medium on the effective refractive indices of the modes propagating through the taper, which can be probed interferometrically, as described above, or by using mode coupling devices such as in-fibre gratings [27], surface plasmon resonances and lossy mode resonances [28].

Optical interferometry offers highly sensitive platforms for sensor development. The non-adiabatic tapered optical fibres described in the previous section are an example of a modal interferometer (MI). Other approaches include in-fibre gratings, regions of core mismatch and directional couplers [29]. Optical fibre MI rely upon the creation of a region where energy is coupled into two modes of the fibre. The modes then propagate along the sensing region of the fibre with different effective refractive indices and respond to the measurand differently.

Fibre optic MI are based on the interference between higher order and fundamental modes, which leads to a sinusoidal channelled transmission spectrum containing fringes described in section 1.2. The phase of the fringes of the spectrum depends on the difference in the optical path lengths of the interfering modes, according to [29]:

$$\begin{eqnarray}&&\varphi =\displaystyle \frac{2\pi }{\lambda }(\delta {n}_{{\rm{e}}{\rm{f}}{\rm{f}}})L,\end{eqnarray} \tag{ 3 }$$

where λ represents the wavelength, L is the centre-to-centre distance between two coupling elements and δn_eff is the difference in refractive indices between the higher order and fundamental modes [29]. As the evanescent fields of modes of different order extend different distances into the surrounding medium, changes in the optical properties of the medium will influence the modes' effective refractive indices differentially, causing changes in the phase of the MI and of the channelled spectrum.

MI with configurations akin to the Mach–Zehnder interferometer (MZI) and Michelson interferometer (MchI) can be implemented in optical fibres. In the context of optical fibre tapers, as discussed in section 2.2, an MZI can be realized by fabricating a non-adiabatic taper of diameter <10 μm, in which two modes are excited at the first taper transition, which propagate with different effective refractive indices and which interfere at the second taper transition. External stimuli cause changes in the effective indices of the modes and thus induce a phase change in the channelled spectrum, figure 4(b) [26, 30]. An optical fibre MZI can also be created between two identical non-adiabatic tapers as shown in figure 5 [29]. The first taper couples the core mode to higher order cladding modes, while second recombines these modes to interfere and create the channelled spectrum. Changes in environmental parameters, such as temperature or RI, around the section of fibre separating the two tapered regions will induce a phase change that will result in a shift of the interference features of the spectrum. Optical fibre MZIs can also be implemented using coupling elements in the form of in-fibre gratings, regions of core mismatch and directional couplers [29].

Zoom In Zoom Out Reset image size

In an example of an optical fibre MchI, illustrated in figure 6, the non-adiabatic taper couples the core mode to higher order cladding modes, which are reflected at the gold coated tip of the optical fibre and are recombined at the same taper to produce interference fringes. Again, changes in environmental around the section of optical fibre separating the taper and the gold coated tip will induce a phase change that will result in shift of the interference features of channelled spectrum. For the same physical length of the interferometer, a MchI will exhibit higher sensitivity as light traverses the sensing region twice. MchI can also be implemented using coupling elements in form of in-fibre gratings and regions of core mismatch [29].

Zoom In Zoom Out Reset image size

These types of interferometers can be interrogated by measuring the wavelength shifts of the features in the channelled spectrum or by recording an amplitude change at a fixed wavelength, table 1. Another example of an interferometric OFS is the so called Fabry–Perot interferometer (FPI), which consists of a cavity formed between two reflectors, as illustrated in figure 7. This cavity can be formed by a gap between two fibres, or between the tip of the fibre and a membrane (figure 7(a)). Alternatively, a thin film can be deposited at the tip of the fibre to form an FPI, where the interface between the fibre tip and the film forms one reflector and the interface between the film and the surrounding medium forms the second mirror.

Zoom In Zoom Out Reset image size

The total reflected power (P_R) for thin film FPI case is described by [29]:

$$\begin{eqnarray}&&{P}_{R}={R}_{1}+{(1-A)}^{2}{(1-\alpha )}^{2}{(1-{R}_{1})}^{2}\cdot {R}_{2}\cdot {e}^{-4\alpha L}\\ &&\,\,+\,2(1-A)(1-\alpha )(1-{R}_{1})\sqrt{{R}_{1}{R}_{2}}\\ &&\,\,\cdot {e}^{-2\alpha L}\,\cos \left(\displaystyle \frac{4\pi {\eta }_{f}L}{\lambda }\right),\end{eqnarray} \tag{ 4 }$$

where R₁ = [η_fibre − η_f/η_fibre + η_f]² is the Fresnel coefficient at the interface fibre-film and R₂ = [η_f − η_air/η_f + η_air]² is the Fresnel coefficient at the interface film-air; α is the transmission loss factor inside the cavity (absorbance) and A is the loss coefficient at the first interface fibre-film; η_fibre is the RI of the fibre, η_air is the RI of the air, L is the physical thickness of the nano-coating and η_f is the effective RI of the film. A change in either the RI of the cavity or its length will modulate the reflected power, which can be used to detect the measurand that modulates the coating's parameters.

Fibre optic devices based on grating structures inscribed into optical fibres have been investigated extensively for sensing applications [31]. An optical fibre grating consists of a periodic modulation of the propagation constants of the modes of the optical fibre, typically induced by a modulation of the RI of the core of the fibre by exposure to a spatially modulated intensity pattern from a UV or femtosecond laser [31]. In the case of a grating that has a period on the order of the wavelength of light (around 1 μm), a so called fibre Bragg grating (FBG), the grating acts to couple the forward propagating mode to a backward propagating mode of the core at a wavelength that satisfies the Bragg condition, which is dependent on the period of the grating and the effective RI of the propagating mode, figure 8. An FBG is viewed as a wavelength selective mirror, which reflects a narrow band of wavelengths back along the optical fibre while transmitting all other wavelengths. The reflected Bragg wavelength is governed by [31, 32]:

$$\begin{eqnarray}&&{\lambda }_{{\rm{Bragg}}}=2{\eta }_{{\rm{eff}}}{\rm{\Lambda }},\end{eqnarray} \tag{ 5 }$$

where n_eff is effective RI of the mode propagating in the core and Λ is the period of the grating.

Zoom In Zoom Out Reset image size

The Bragg wavelength is sensitive to parameters that influence the period of the grating and RI of the fibre, typically strain and temperature [31, 32]. The wavelength-encoded nature of the sensor offers a number of advantages, not least the ability to multiplex a serial array of FBG sensors in a single optical fibre by ensuring that each has a different period, and thus a different Bragg wavelength. To sensitize the Bragg wavelength to the surrounding RI, the fibre has to be polished, etched or tapered to allow the EW to interact with the surrounding medium. The interaction of the EW with the surrounding medium influences the effective index of the propagating mode, resulting in changes in the Bragg wavelength, and, as the attenuation of the taper is dependent upon the surrounding RI, changes in the reflected power. The total wavelength shift (Δλ) due to strain (ε) and temperature change (ΔT) can be approximated by equation (6) [31, 32].

$$\begin{eqnarray}&&{\rm{\Delta }}\lambda ={\lambda }_{{\rm{Bragg}}}[(1-{P}_{e})\varepsilon +((1-{P}_{e})\cdot \alpha +\xi ){\rm{\Delta }}T],\end{eqnarray} \tag{ 6 }$$

where P_e the photoelastic is constant, ξ is the thermo-optic coefficient of the optical fibre and α is the thermal expansion coefficient [31, 32].

Tilted FBGs (TFBGs), where the grating planes fabricated such that they are inclined with respect to the axis of the fibre, can be used to measure the changes in the surrounding RI without the need for further modification of the fibre. TFBGs, in addition to coupling light to a backward propagating mode at the Bragg wavelength, also couple light to forward propagating modes of the cladding, creating further spectral features. The interaction of the cladding modes' EWs with the surrounding medium changes their effective refractive indices, resulting in changes to the wavelengths of the cladding mode coupling features (the Bragg wavelength is unchanged) [33].

In contrast to FBGs, a long period grating (LPG) typically has a period in the range 100 μm to 1 mm and couples light from the core mode to a discrete set of co-propagating cladding modes, also shown in figure 8 [13]. As the cladding modes suffer high attenuation, the transmission spectrum of an LPG is characterized by a series of resonance bands, each corresponding to coupling to a different cladding mode. The resonance wavelengths are dependent on the product of the period of the LPG and the difference between the core and cladding mode indices, governed by [34]:

$$\begin{eqnarray}&&{\lambda }_{i}=({n}_{c}-{n}_{{\rm{clad}}}^{i}){\rm{\Lambda }},\end{eqnarray} \tag{ 7 }$$

where Λ is the period of the grating, n_c is the RI of the mode propagating in the core, and ${n}_{{\rm{clad}}}^{i}$ is the effective index of the ith axially symmetric and linearly polarized cladding mode [34, 35].

As a result of the interaction of the EW of the cladding modes with the surrounding medium, the resonance wavelengths are inherently sensitive to changes in the surrounding RI, and to the optical thickness of nanoscale coatings deposited onto the cladding. This makes LPG sensors ideal candidates for development of refractometers and biosensors. In biosensor applications, since the LPG is sensitive to both bulk (due to concertation change) and surface (due to binding events) RIs, special care needs to be taken to address this difference. The different ways of addressing this and enhancing the sensitivity of LPGs in bio-sensing applications have been thoroughly described in recent paper by Chiavaioli et al [3].

LPGs fabricated in standard single mode optical fibres offer inherent sensitivity to a number of parameters, predominantly strain, temperature, curvature and surrounding RI [34]. The temperature sensitivity of the central wavelength (λ_i) of the LPG can be calculated using equation (8) [36] and the sensitivity to the variation of the surrounding RI (n₃) can be calculated using equation (9) [37],

$$\begin{eqnarray}&&\displaystyle \frac{d{\lambda }_{i}}{dT}=\displaystyle \frac{d{\lambda }_{i}}{d(\delta {n}_{{\rm{e}}{\rm{f}}{\rm{f}}})}\left(\displaystyle \frac{d{n}_{c}}{dT}-\displaystyle \frac{d{n}_{{\rm{clad}}}^{i}}{dT}\right)+{\rm{\Lambda }}\displaystyle \frac{d{\lambda }_{i}}{d{\rm{\Lambda }}}\cdot \displaystyle \frac{1}{L}\displaystyle \frac{dL}{dT}\end{eqnarray} \tag{ 8 }$$

$$\begin{eqnarray}&&\displaystyle \frac{d{\lambda }_{i}}{d{n}_{3}}=\displaystyle \frac{d{\lambda }_{i}}{d{n}_{{\rm{clad}}}^{i}}\cdot \displaystyle \frac{d{n}_{{\rm{clad}}}^{i}}{d{n}_{3}},\end{eqnarray} \tag{ 9 }$$

where L is the length of the LPG, Λ its period, $\delta {n}_{{\rm{e}}{\rm{f}}{\rm{f}}}={n}_{c}-{n}_{{\rm{clad}}}^{i}$ (differential effective index). The cross-sensitivity to strain is not considered in this treatment since material (strain-optic and Poisson's effect) and waveguide (change of the grating period) contributions to strain sensitivity can cancel each other out having opposite signs [36, 37]. The phase matching condition described by equation (7) contains a turning point at which it has been demonstrated [38–40] that the sensitivity of the attenuation band to perturbation of the surrounding environment is at its maximum, i.e. there is optimum sensitivity when the LPG is fabricated with a period such that the coupling of light to the cladding mode occurs near to or at the phase matching turning point [41]. Among the various types of OFS, those based on optical fibre gratings offer wavelength-encoded information that overcomes the referencing issues associated with intensity based approaches.

The properties of the OFS described in this section are summarized in table 1. According to the measurands, OFS can be broadly classified into physical, chemical and biosensors.

As mentioned in [3], it is often difficult to compare the performance of the different OFS even when same measurand is measured, as authors usually provide limited information, typically stating only the units of sensitivity, which differ when different principles of operation of the sensors is employed (table 1). For example, sensitivity to temperature measured using an FBG will be reported in pm °C⁻¹ while that measured by an interferometer be reported in rad °C⁻¹ or dB °C⁻¹, which makes comparison of sensors in terms of sensitivity meaningless. There is a need for the optical sensor community to develop standardized units of measurements to allow comparison of the performances of different types of OFS. Perhaps the performance of the sensor can be expressed in terms of a figure of merit that will allow this comparison—this could well be the range, resolution/accuracy or limit of detection (LoD) of the particular measurand, as this takes into account standard deviation. This discussion, however, is beyond the topic of the current review paper and readers are referred to [3] for a detailed discussion of OFS metrology.

Temperature is one of the vital signs and is a crucial and routinely monitored parameter in medicine that is measured using a variety of technologies [42] in all clinical settings, including surgeries, oncology treatment and intensive care units [43]. In healthcare, the temperature sensing requirements are application dependent, but generally a temperature range of 35 °C–45 °C with a resolution of at least 0.1 °C is required [44]. The required response time of the temperature sensor is also application dependent [2]. For some thermal treatment procedures, such as high-intensity focused ultrasound ablation, the coagulative temperature (43.5 °C–57.0 °C) is reached within less than 30 s [45], while for laser ablation this can lie between 5 and 15 min [45].

Typically, OFS will be deployed to measure temperature when immunity to electromagnetic interference or electrical insulation are required [44, 46], such as during magnetic resonance imaging (MRI) or radio frequency (RF) treatment [46–48]. OFS temperature sensors exploit a range of transduction principles, such as temperature dependent fluorescence lifetime [49–52], Rayleigh scattering (change in the amplitude of the back-reflected signal with temperature) [53] and thermal expansion and the thermo-optic effect in FBGs [54–56], LPGs and FPIs [57, 58]. Table 2 summarizes key parameters of optical fibre temperature sensors [2].

In addition, optical fibre temperature sensors, especially those based on gratings and Rayleigh scattering, offer a prospect for multiplexed or distributed temperature sensing, allowing the generation of a heat map of the area of interest [27, 62, 64, 65]. Figure 9 shows an example of the temperature distribution measured inside liver tissue (figure 9(a)) using a Luna OBR4600 optical fibre distributed temperature sensor (DTS), measuring changes in the Rayleigh back scattering signal in the optical fibre, during an RF ablation procedure [62]. The equipment was configured such that measurement time of 1 s for the entire fibre length (<20 m) and choosing an accuracy estimated as 0.5 °C, a minimum spatial resolution of 200 μm in real time (and 20 μm in post-analysis with data processing) was achieved. The performance of the DTS Luna Rayleigh back scattering system depends on a tight trade-off between accuracy, spatial resolution, and sampling time. Analysis of the thermal maps verified that the spatial and temporal evolutions of the temperature were smooth and consistent with what would be physically anticipated [62]. The authors reported that measurements can be employed to investigate the physics of RF ablation and for the validation of ablation models. Results can be used for investigating the physics of thermal ablation, to set the basis for distributed temperature monitoring and for the validation of models [62]. It has the potential to facilitate the adjustment of treatment to the specific biological and electrical properties of the tissue undergoing ablation.

Zoom In Zoom Out Reset image size

Despite the commercial availability of OFS systems targeted at healthcare, based on various measurement principles, offered by FISO, LumaSence, Neoptix and OpSens, there are limited reports of in vivo use of OFS with animals or humans [2]. Fajkus et al [66] proposed a system based on FBGs for vital sign measurements (body temperature, heart rate and respiratory rate). The system consisted of 32 FBG sensors that were embedded into polymer to increase their temperature sensitivity. The FBGs were interrogated using a wide-spectrum SLED (1512.5–1587.5 nm and an output power of 1 mW) connected to spectrum analyser via a fibre optic circulator. The system was tested on 10 human volunteers, with the measurements showing good correlation with standard equipment routinely used in hospitals [66]. The main advantage of the system identified by authors was the design of the novel patient-friendly non-invasive measurement probe.

For applications such as cryo-ablation, OFS need to perform at temperatures down to −180 °C [67]. Samset et al [67] measured the temperature distribution using 10 wavelength division multiplexed FBG sensors. Adjacent sensor elements were separated by 6.5 mm and the external diameter of the probe was 1.25 mm. The median difference between measured and estimated temperature was 3.03 °C, and the measurements provided valuable information for the development of models of cryo-ablation. Temperature maps as outlined in the work may be used for monitoring of cryotherapy in order to increase clinical effectiveness.

Yoo et al [68] proposed respiration sensors based on thermo-chromic material deposited onto the tip of plastic optical fibre for respiratory monitoring inside an MRI system. Two types of non-invasive fibre optic respiration sensors were reported, one being a nasal-cavity attached sensor that could measure the temperature variation of air-flow using a thermo-chromic pigment. The other was an abdomen-attached sensor that measured the abdominal circumference change using a sensing component composed of polymethyl-methacrylate tubes, a mirror and a spring. The intensity of reflected light was changed by the variation of the distance between the mirror and the distal end of the plastic optical fibre according to abdominal movement. The authors verified that respiratory signals can be obtained with no detrimental effects on the magnetic resonance image.

Musolino et al [69] described a fluorescence based optical fibre temperature sensor based on rear earth glass deposited onto the tip of an optical fibre. The sensitivity was reported to be 0.005 258 K⁻¹ over a temperature range of 22 °C–51 °C. The sensor was tested successfully in vivo by measuring the brain temperature of rats [69]. A comprehensive review of optical fibre temperature sensors and their applications in healthcare can be found in [2].

The temperature sensitivity of the light propagating through optical fibres compromises the sensing of other measurands, and hence temperature needs to be taken into account when measuring other parameters. This can be done either by developing temperature insensitive sensors [70, 71] or by measuring temperature independently to other measurand and subsequently compensating for its effect [72].

Pressure in the human body is grouped into two categories, isotropic pressure, such as blood pressure and intracranial fluid pressure, and contact pressure such as intra vertebral pressure. Blood pressure, in particular, is a vital sign in the human body that is routinely measured during medical examinations and procedures [1, 73]. The pressure range of interest, however, is very wide, is dependent on the part of the body where the measurement is performed and is influenced both by external and internal factors [73]. The pressure in the human body at normal physiological conditions is typically divided into three levels [73, 74]:

• low pressure domain—between 0 and 10 mbar (1 mbar =100 Pa ≈ 0.75 mmHg); (capillaries, brain, urinary bladder, and muscular compartments)
• medium pressure domain—the values lie between 25 and 150 mbar (circulatory system including the heart)
• high pressure domain—the values can be as high as 180 mbar (load bearing structures like hips and knees) (135 mmHg).

At pathophysiological conditions, the pressure can vary significantly and range from –0.1 bar (–10 kPa, –75 mmHg) for intra-alveolar and intra-tracheal pressure to as high as 0.4 bar (40 kPa, 300 mmHg) for aortic and left ventricular pressure [74]. These requirements place constraints on optical fibre pressure sensors. Ideally, an optical fibre pressure sensor should be able to operate between 0.04 bar (–4 kPa, –30 mmHg) in the heart [75] to 10 bar (7500 mmHg) in bones and joints [76, 77] with a typical resolution of 1 mbar (0.75 mmHg) and accuracy of <1% [44]. In practice, however, the sensors will cover much narrower range to satisfy requirements of a particular application.

A variety of OFS can be used to measure pressure, such as intensity modulated, FBGs [78–80] and FPIs [81].

Intensity modulated pressure sensors were originally proposed for pressure measurements in 1960s [82] but, despite the simplicity of the design and its cost effectiveness, they have not found widespread application. There are two main designs, one based upon a movable reflectance membrane interfaced with the tip of the optical fibre such that the intensity of light reflected back into the fibre depends on the pressure [83, 84]. The second relies upon bending losses, where the pressure dependent bend will induce loss in light propagation inside the optical fibre [85]. The main drawback of these approaches is their reliance on the measurement of intensity, which can be influenced by a wide range of parameters including light source fluctuation, changes in the reflectivity of the membrane (due to temperature or oxidation effects) and bending of the fibre [86].

The inherent sensitivity of single mode silica optical fibre FBGs to hydrostatic pressure is not very high (0.304 pm bar⁻¹ (4.05 × 10⁻⁴ pm mmHg⁻¹)) [87]. Increasing the pressure sensitivity is typically achieved by transducing the pressure into a strain acting along the axis of the fibre by appropriate packaging. One of the approaches consists of attaching the FBG to a flexible diaphragm either orthogonally or in the diaphragm plane in areas where the strain is maximal. Another approach involved the packaging of the FBG sensor in a cylindrical assembly so that pressure sensitivity can be increased via mechanical amplification schemes [86]. Sensitivity can be enhanced to 270 pm bar⁻¹ (0.36 pm mmHg⁻¹) when the FBG is embedded into a flexible patch that translates the transverse load into an axial strain measured by the FBG [88, 89].

Fabry–Perot sensors offer the highest sensitivity to pressure and can be implemented with great design flexibility [86]. A compact pressure sensor was developed by FISO [90], where the diameter of the sensor head was 125 μm with standard range of ±300 mmHg and system accuracy of ±3 mmHg. A detailed review of optical fibre pressure sensors is beyond the scope of this paper, but more details of optical fibre pressure sensors can be found in recent reviews in [77, 81, 91]. Table 3 provides key parameters of optical fibre pressure sensors [91].

Optical fibre pressure sensors have been trialled in both ex vivo and in vivo applications in various areas of healthcare, including gastroenterology, cardiology, neurology and urodynamic. The review by Poeggel et al [1] summarizes comprehensively these applications. Optical fibres have also been used for contact pressure sensing in applications in biomechanics and rehabilitation or at the interface between the medical device and tissue (e.g. sockets, bandages [77], in bone, dental biomechanics and as insole and amputee socket sensors). A number of applications were reviewed by Poeggel et al [1] and by Al-Fakih [77]. Here, more recent examples of in vivo pressure sensors applications are outlined.

OFS were applied recently to address an important problem known as medical device related pressure ulcers, which causes over a third of soft tissue damage in hospitals [98, 99]. Here, the optical fibre pressure sensor was based on an FBG embedded into a polymer patch with high sensitivity to contact pressure, which could be used to measure at the interface between a medical device and tissue [89, 100].

The Arkwright group [101, 102] has recently proposed an interesting approach to pressure sensing involving the use of FBG sensors entwined in a double helix, which allows real time, distributed sensing of pressure to be performed with inherent temperature immunity. The sensor can be used in compression bandaging and for in vivo colonic diagnostics [80]. Compression bandaging is a major approach in healthcare for the treatment of conditions such as venous leg ulcers and lymphoedema, where precise control of the applied pressure is essential. The FBG based sensors were blind tested on 46 humans with results showing a correlation to the target gradient of 93% [102].

One example of internal tissue damage that can occur is post-intubation laryngotracheal stenosis, which is caused by use of an endotracheal tube (ETT) during mechanical ventilation. It is a life threatening debilitating condition with a protracted course, typically requiring multiple complex surgical procedures. ETT is used in mechanical ventilations and consists of lumen that is used to deliver artificial air and an inflatable cuff that holds tube in place and ensures that there is no leak of secretions into the lungs; figure 10. A sensorised ETT was developed with the aim of aiding a reduction in soft tissue damage and improving design of future devices. Optical fibre sensing was used to monitor both cuff contact pressure (via a packaged FBG) and tissue perfusion (via reflectance photoplethysmography (PPG), which detects blood volume changes in the microvascular bed of tissue). The monitoring system was used successfully in an animal study, demonstrating reliable measurement of contact pressure and perfusion (PPG) [100]. Both the FBG and the perfusion sensor were bonded to the outside of the cuff and an additional outer cuff was bonded to the ETT in order to enclose the sensors and inner cuff (figure 10).

Zoom In Zoom Out Reset image size

The FBG sensor incorporated into the ETT was interrogated using a commercially available system (Smartscan, Smart Fibres Ltd, Bracknell, UK), which uses a tuneable laser diode to provide swept wavelength illumination of the FBG. This wavelength shift is related to the applied pressure after compensating for temperature by subtracting readings from an FBG that was not embedded into the patch. Perfusion was measured using reflectance PPG and the recorded data sent to a laptop for further processing. The device was tested on six pigs to investigate whether pressure and blood flow could be monitored reliably in vivo. The results are shown in figure 11. It is interesting to note that the FBG could measure changes in respiration due changes in the contact pressure between the sensor and trachea (figure 11(a)). The measurements also revealed that high cuff pressures occluded the blood flow, such that the pulsatile signal could not be observed at high pressure but returned when pressure was reduced. The sensor demonstrated high quality reliable PPG signal return from tissue when the cuff was brought into optimum contact with the trachea.

Zoom In Zoom Out Reset image size

The response of an FBG to strain can be used to measure the periodic mechanical movement of the chest wall caused by breathing or cardiac contractions and hence monitor respiration and heart rate [103]. Heart rate can be measured using FBG sensors via the detection of the pressure pulse waveform (PPW) generated by the heart that travels through the human body [104]. The principle of operation of such a sensor is based on the detection of the change of the Bragg wavelength of an FBG placed close to an artery by the PPW travelling in the artery. A number of configurations have been demonstrated, including adhering the FBG to a silicone diaphragm, which in turn is adhered to a hollow box [105], a portable pen-like OFS [106], and a lever amplification mechanism [104].

Similar principles may be used to measure respiratory rate, where movement of the chest generates a strain acting on FBG [103] or LPG [107, 108] sensors. Both LPGs and FBGs can be embedded into textiles to create wearable unobtrusive sensor devices [109, 110]. OFS have been used extensively in monitoring respiration in MRI scanners and a useful review is provided in [111].

The other physical parameter that is routinely measured in clinical settings is blood flow [112]. Optical fibre Doppler flowmetry is used routinely for blood flow measurements [113]. The role of the optical fibre in this measurement is to deliver to, and collect light from, the measurement point. Optical fibre velocimeters have been reviewed in detail in [114]. The role of the optical fibre in these measurements is simply to deliver and collect light to and from the measurements point and will not be discussed in further in this review.

The detection of chemical compounds in the human body can be conducted in the gas phase, analysing exhaled gases in the breath or gases emitted through the skin, or in the liquid phase, in samples such as urine, saliva, sweat, tears and blood.

4.1.1. Detection in gas phase

Gaseous compounds excreted from the human body (biomarkers) reflect certain metabolic conditions as well as the blood gaseous content, which offers a possibility for the development of non-invasive diagnostics [122]. The classification of bacteria related to human diseases [123, 124], urinary tract infections [125] and further progress to metabolic disorders such as diabetes [126] or renal dysfunction [127] has been achieved by monitoring biomarkers. Further, renal failure in rats [128] and lung cancer detection in people [129] have been achieved using the breath sniffing method and using arrays of appropriately modified chemo-resistors.

Despite the interest and success in characterizing human odours, only few methods have the potential to be employed in clinical practice. Most of these methods have been used for monitoring the presence of gases such as CO, NO, isoprene [130], acetone [131], and ammonia [132] in human breath.

CO₂ and oxygen are two gases that are routinely measured in clinical settings [133]. OFS can be used to measure both of these gases via the simultaneous excitation of two fluorescent pH sensitive indicators with well-separated emission bands [134]. It was reported that oxygen can be continuously monitored in the 0–200 Torr (0–26.6 kPa) range with ±1 Torr accuracy and that CO₂ can be monitored in the 0–150 Torr (0–20 kPa) range with ±1 Torr accuracy. The accuracy is higher at low partial pressure, so that the detection limits are defined at ∼0.5 Torr in both cases [134].

In clinical settings, IR spectroscopy (capnography) is used to detect CO₂. In demonstrations of CO₂ OFS, pH sensitive indicators are usually used. Kole et al measured breath-to-breath CO₂ and oxygen using an optochemical oxygen sensor, exploiting the measurement of the quenching of the intensity of photoluminescence of the dye platinum (U)-octaethylporphyrin-ketone [135]. One of the issues with dye based OFS is the lifetime, as the dye tends to leach and thus the coating loses its sensitivity. An optical fibre CO2 sensor based on an LPG coated with metal organic compounds (HKUST-1) to detect CO2 was proposed recently [136]. The sensor operated in the range of 500–40 000 ppm of CO₂ with the LoD of 401 ppm.

Ammonia is one of the major metabolic compounds and the importance of its detection has been emphasized recently because of its correlation with specific diseases such as dysfunction in the kidney and liver [137–141]. At normal physiological conditions, ammonia can be expelled from the slightly alkaline blood and emanated through the skin or exhaled with the breath. The detection of the ammonia present in breath or urine can be used for the early diagnosis of liver or stomach diseases [138]. The development of sensing devices for measuring ammonia with a sensitivity of 50–2000 ppb and with a fast response time is highly desired [137].

To the best of our knowledge, there is still no OFS that is capable of measuring ammonia in the breath. Conventional methods for the ammonia measurements are mainly based on gas chromatography mass spectrometry, which, in spite of its high selectivity and sensitivity, is expensive, requires well trained operator and is time consuming. The development of cheap, small, sensitive and reliable sensor devices that can efficiently operate at different relative humidity (RH) levels can help to create a point-care medical system that can be used in daily life.

In past decades, a number of approaches to optical fibre based ammonia sensing utilizing sensitive coatings have been reported. Wolfbeis employed fluorescent measurements using optical fibre probe to measure pH of an indicator solution [142]. Since then, lossy mode resonances [143], EW [144–146] fibre gratings and reflection sensors at the tip of the fibre [147] have been reported. Generally, the LoD ranges from 10 to 100 s of ppm, which is too high for meaningful healthcare application. Table 4 summarizes some of the ammonia sensor parameters reported in the literature.

Table 4.  Summary of optical fibre based ammonia sensor parameters.

Sensor type	Sensitive element	Limit of detection (LoD)/lowest measured concentration (LMC)	Response time	Reference
Cladding removed evanescent wave	Universal pH indicator	10 ppm(LMC)	5 min	[144]
	Bromocresol purple/bromocresol green, dip coating sol-gel	9 ppm (LMC) 0.014 dB ppm−1	8 s	[145]
	Bromocresol purple, sol-gel	145 ppm (LMC)	10 s	[146]
	TSPP	6 ppm	15 s	[13]
Reflection type	ZrO2/PSS Nano-assembled	1 wt% (LMC)	Minutes	[148]
	Oxazine 170 perchlorate	200 ppm (LMC)	—	[149]
Lossy mode resonance	Titanium dioxide containing TMPyP	0.1 ppm (LMC)	30 s	[143]
Grating based	PDDA/TSPP	0.67 ppm (LoD)	—	[147]

Although the detection of VOCs is not routinely conducted in clinical settings, it has high potential as a non-invasive diagnostic tool and much literature has been published on the relationship between diseases at the VOCs emitted from human body [150]. A sensor based on the cladding removed EW OFS modified with sensitive layer of dye compound that was tested on several humans was able to distinguish skin emanations from different people was reported in [151]. Using this sensor it was also possible to recognize changes in people's physiological conditions by applying a pattern recognition technique (principal component analysis) to analyse changes in the EW spectrum. The developed system has a significant advantage in size, simplicity of fabrication and cost as compared to the arrays of chemical sensors or more sophisticated gas chromatography techniques. Being able to recognize qualitative changes in 'global body odour', the described system has the potential for further development for application in healthcare for disease diagnostics.

Several OFS have been used to detect VOCs, although none yet have sufficient sensitivity and selectivity for meaningful medical application [152]. There have been a number of fibre optic based VOC gas sensors proposed recently using the deposition of a sensitive layer on the fibre tip to form an FPI [153–157], or cladding removed EW sensors, where part of the cladding is mechanically or chemically removed and then the functional coating is deposited over the stripped region [158–162]. Coatings sensitive to VOC vapours have been also deposited onto the surface of photonic crystal fibres [163] and LPGs [164–166]. The reported sensors suffer mostly from a long response time, in the range of 10 s of minutes, and are usually tested over a limited but biologically relevant concentration range.

Humidity measurement plays an important role in the monitoring of air delivered to mechanically ventilated patients in critical care [167] and in monitoring the microenvironment of chronic wounds [168]. Humidification of inspired gases is an essential part of clinical treatment in critical respiratory care. Inhalation of inadequately humidified gas during invasive (through an ETT placed in the patient's trachea) or non-invasive (via a mask placed on the patient's face) ventilation, causes drying of the delicate respiratory mucosa and consequent cooling, mucosal injury, drying of (and difficulty in clearing) secretions and respiratory deterioration [169]. The clinically acceptable range of absolute humidity (AH) and RH values at the level of the upper trachea is between 5 mg l⁻¹ (50% RH at 27 °C–28 °C) and 42 mg l⁻¹ (85% RH at 34 °C nasal, 95% RH at 35 °C naso/oropharynx) [170]. Requirements for response and recovery times of humidity sensors are driven by the respiration rate and typical need to be faster than 1 s [167].

A number of OFS configurations have been used to measure humidity, such as FPI using various humidity sensitive thin films [167, 171–175], de-clad plastic optical fibre [176, 177], hetero-core OFS [178] and microstructured optical fibres [179]. FBGs modified with materials that swell and contract when exposed to water can also be used to measure humidity [180], but this type of sensor possesses relatively slow response and recovery times and is thus not suitable for medical application.

The principle of operation of FPI humidity sensors is based on the detection of the RI change caused by the absorption of water molecules into a humidity sensitive film, which can provide fast response to changing humidity levels [167]. An FPI humidity sensor was reported recently that was tested within a mechanical ventilator to detect respiration rate, see figure 12 [167], and to determine the AH delivered into the lungs [181]. The response time was 1.5 s. The sensor has also been validated with animal models [100, 182].

Zoom In Zoom Out Reset image size

Another application of humidity sensing is in wound healing [177]. A wound dressing that remotely monitors parameters associated with healing such as the humidity presented on the wound microenvironment, wound exudate pH and proliferative cell levels within the wound could have a significant impact on wound treatment. This monitoring would enable clinical interventions to take place promptly but only when required, thus improving wound care and reducing the number of clinical appointments [183]. The humidity sensor had sensitivity of approximately −3.87 × 10^–3 and −9.61 × 10⁻³ in transmittance percentage per RH percentage for the range of ∼10% to ∼75% RH and 90% to 97% RH, respectively. The proof of concept measurements made on the skin using a sensor embedded into a wound dressing, see figure 13, indicated that this sensor has the potential to be used to monitor the humidity of the skin microenvironment within a wound dressing, which can be used to provide better prognosis of healing.

Zoom In Zoom Out Reset image size

4.1.2. Detection in liquid phase

The commonly accepted classical definition of a biosensor is [199]: 'Biosensors are a particular class of devices, where the sensing unit is constituted by a biological recognition element, able to recognize in a highly specific manner a selected target or a class of compounds'. The key in this definition is 'biological recognition element' and therefore, as discussed by Wang and Wolfbeis [200], 'chemical sensors not using a biological component but placed in a biological matrix (such as a pH electrode in blood) are not biosensors by definition'. Recently, Socorro and Diaz [201] suggested that the biosensor definition should be updated and expanded to include 'devices capable of monitoring a specific biomedical variable (including biochemical or physical) and provide information that can give a rapid and accurate diagnosis of a patient's health condition'. Although this is an interesting suggestion, perhaps this definition is more suitable for a healthcare sensor rather than a biosensor and therefore in this review we will follow a classical definition of biosensor as a device with a biological recognition element. This biological recognition element, typically bound to an optical fibre via covalent bonding, could be an antibody [40, 202], an enzyme [203], a protein [204], or a nucleic acid [205] or even a whole cells [206, 207], chosen to recognize analytes via biochemical mechanisms.

The sensing system can be designed to operate in labelled and label-free mode. The former utilizes fluorescence labelling of the analyte and measures changes in fluorescence intensity caused by the interaction of analyte with the biological recognition element [202]. This approach is similar to enzyme-linked immunosorbent assay, ELISA, which is used routinely in healthcare [208]. In the development of optical fibre biosensors, there is more research effort devoted to the label-free mode, since it overcomes the disadvantages of the fluorescence methods, such as the additional labelling step and problems associated with fluorophores, for example their relatively short lifetime and photobleaching. The label-free approach is usually based on the detection of changes in optical thickness (product of RI and geometrical thickness) associated with the binding of analyte to the biological recognition element deposited onto an optical fibre. Optical fibre refractometers based on sensing platforms such as tapered optical fibres [209–213], LPGs [40, 214–220] and TFBG [221] coated with biological recognition elements have been explored for label-free bio-sensing.

Another interesting approach for label-free bio-sensing is based on so called plasmonic OFS [222]. This approach utilizes noble metals, typically gold or silver, (in the form of a film or nanoparticles) grafted with bioreceptors and deposited onto the surface of the optical fibre. The EW is used to excite plasmons (oscillation of electrons at the metal–dielectric interface), which are sensitive to the binding of analytes. One of the biggest advantages of plasmonic OFS over other label-free methods is their high sensitivity [223].

In healthcare, biosensors can be used to detect biomarkers or to detect bacteria directly. Albert et al [223] reported a plasmonic OFS based on a TFBG coated with a thrombin binding aptamer, figure 14(a), to detect the protein human α-thrombin, an important enzyme in the blood-clotting cascade, in its fibrinogen-binding exosite. The authors demonstrated that the specific interaction between the protein and the sensor lead to a signal increase after each new higher concentration, and that an increase of the SPR signal remained after a final protein buffer wash, revealing the binding of the analyte to the sensitive layer, figure 14(b).

Zoom In Zoom Out Reset image size

The authors also proposed a concept for multiplexed TFBG-SPR sensors embedded across a microfluidic channel; figure 15 [223]. This concept allows application of several OFS in a multiplexed array, where the fibres were interrogated sequentially. Each sensor could have its surface functionalised with a different bio-receptor to capture different targets, to provide baseline reference data, or to improve the LOD through statistical analysis of the response of multiple identical sensors.

Zoom In Zoom Out Reset image size

In a similar approach [205], an LPG was functionalised with an aptamer to detect the outer membrane proteins of Escherichia coli in the clinically relevant concentration range between 0.1 and 10 nM. The principle of operation is based on the measurement of the wavelength shift induced by the binding of the membrane protein. The sensor could be regenerated by treatment with acid. An example of an enzymatic optical fibre biosensor, where a particular enzyme is used to detect analyte, is a glucose sensor, which was reported in [224]. In glucose biosensors, glucose oxidase (GOx) is attached to the optical fibre and the change of RI due to the oxidation of glucose by GOx is measured. The sensor operated in the 0–2.6 g l⁻¹ concentration range and was tested in whole blood.

Comprehensive reviews of optical fibre chemical and biosensors have been recently published [3, 199–201, 221]. There are no reports of optical fibre chemical or biosensors that have been used in clinical settings. Moreover, there are no publications that have validated optical fibre chemical and biosensors in vivo, with all published work stopping at the proof of concept stage in the laboratory setting. The possible reasons for this and also some suggestions how this can be overcome are discussed in the next section.